Volume 17, Issue 1, Ver. 2 (Jan – Feb. 2015)

1. IOSR Journal of Computer Engineering (IOSR-JCE)
e-ISSN: 2278-0661,p-ISSN: 2278-8727, Volume 17, Issue 1, Ver. 2 (Jan – Feb. 2015), PP 01-05
www.iosrjournals.org
DOI: 10.9790/0661-17120105 www.iosrjournals.org 1 | Page
Brain Tumor Segmentation and Extraction of MR Images Based
on Improved Watershed Transform
1
J.Mehena, 2
M. C. Adhikary
1
(Research scholar, Fakir Mohan University, Balasore, Odisha, India)
2
(Reader, Fakir Mohan University, Balasore, Odisha, India)
Abstract: Brain tumor extraction in magnetic resonance imaging (MRI) has becoming an emergent research
area in the field of medical imaging system. Extraction involves detection, localization, tracking, enhancement
and recognition of the tumor from the MR brain images. Brain tumor extraction helps in finding the exact size
and location of tumor. The watershed transform is a popular and has interesting properties that make it useful
for many image segmentation applications. The intuitive description of this transform is quite simple, can be
parallelized and always produces a complete division of the medical images. One of the important drawbacks
associated to the watershed transform is the over segmentation that commonly results in brain images. We
present an improvement to the watershed transform in this paper for the extraction of brain tumor based on
segmentation and morphological operator. The tumor may be benign, pre-malignant or malignant and it needs
medical support for further classification.
Keywords: Brain Tumor, Morphological Operators, MRI, Segmentation, Watershed
I. Introduction
In medical imaging, segmentation of tissues and structures plays a vital role in many image analysis
applications developed for medical diagnosis [1]. Image segmentation helps in diagnosis of brain diseases and
helps in quantitative analysis of MR images such as measuring accurate size and volume of extracted portion.
Exact measurements in brain diagnosis are difficult because of different shapes and sizes of tumor [2, 3].
Treatment plans and evaluation of disease progression of that disease affect specific tissues or structures, lead to
loss and abnormalities. An accurate, reliable, and automatic segmentation of these tissues and structures can
improve diagnosis and treatment of brain diseases [4, 5]. Manual segmentation is bias and usually accurate but
is impractical for large datasets because it is tedious and time consuming. Automatic segmentation techniques
can be useful for clinical applications if they have: ability to segment like an expert, excellent performance for
diverse datasets and reasonable processing speed for large datasets because it is tedious and time Consuming [6,
7].
Brain tumor is one of the major causes for increasing mortality among children and adults. A tumor is a
neoplasm which is formed by an abnormal growth of cells. Brain tumors can be separated into two general
categories depending on the tumors origin, their growth pattern and malignancy. Tumors that arise from cells in
the brain or from the covering of the brain are called primary brain tumors. Tumors occur when cancer cells
spread to the brain from a primary cancer in another part of the body are called secondary brain tumors. It has
been concluded from the research point of view that the number of people suffering and dying from brain
tumors has been increased perhaps as much as 300 over past three decades. The best type of imaging to
diagnose most types of brain tumors is MRI [8, 9]. This technique is basically used to detect the differences in
the tissues which have a far better technique as compared to computed tomography. So this makes this
technique a very special one for the brain tumor extraction. In this paper we purpose an improvement to the
watershed transform for accurately detecting the location of brain tumor in MR images.
This research paper is organized as follows. In Section II, the proposed watershed transform is
described. Experimental results and comparison with existing extraction algorithms are presented in Section III.
Finally, conclusions and discussions come in Section IV.
II. Proposed Technique
The proposed technique for the extraction of brain tumor consists of the following processes, as shown
in Fig.1. Pre-processing, improved watershed transform, morphological operations and resulting in the brain
tumor designate confirmation. We now discuss the above mentioned steps in detail:

2. Brain Tumor Segmentation and Extraction of MR Images Based on Improved Watershed Transform
DOI: 10.9790/0661-17120105 www.iosrjournals.org 2 | Page
Fig.1. Flow Chart of Proposed Technique
2.1 Preprocessing
In this stage MR brain image is enhanced in the way that finer details are improved and noise is
removed from the image. Commonly used enhancement and noise reduction techniques are implemented that
can give best possible results[10]. Enhancement will result in more prominent edges and a sharpened image is
obtained, noise will be reduced thus reducing the blurring effect from the image. In addition to enhancement,
image segmentation will also be applied. This improved and enhanced image will help in detecting edges and
improving the quality of the overall image. Edge detection will lead to finding the exact location of tumor [11].
In this work sobel edge-emphasizing filter is used to sharpen the MR brain image as we need the sharp
edges because this will help us to detect the boundary of the tumor.
2.2 Methodology
2.2.1 Watershed Transform
Watershed transform [2] is a most efficient segmentation technique coming from the field of
mathematical morphology. The intuitive idea of this transform is quite simple: if we consider the image as a
landscape or topographic relief, where the height of each point is directly related to its gray level, and consider
rain gradually falling on the terrain, then the watersheds are the lines that separate the “lakes” actually called
catchment basins that form. The watershed transform is computed on the gradient of the original image, so that
the catchment basin boundaries are located at high gradient points. This transform has been widely used in many
fields of image processing, including medical MR image segmentation, due to the number of advantages that it
possesses: it is quite simple, intuitive, fast, parallelized technique and produces a complete division of the image
in separated regions even if the contrast is poor, thus avoiding the need for any kind of contour joining. Some
important drawbacks associated to the watershed transform are the over segmentation and poor detection of
significant areas with low contrast boundaries that commonly results in MR brain images.
2.2.2 Marker Controlled Watershed Segmentation
Touching objects separation in an image is one of the most difficult works in medical image
processing. The watershed transform is often applied to this problem. The watershed transform finds catchment

3. Brain Tumor Segmentation and Extraction of MR Images Based on Improved Watershed Transform
DOI: 10.9790/0661-17120105 www.iosrjournals.org 3 | Page
basins and watershed ridge lines in an image by treating it as a surface where light pixels are high and dark
pixels are low. Segmentation using the watershed transform works better, if you can identify or mark foreground
objects and background locations. Marker-controlled watershed segmentation follows this basic procedure [11]:
i. Compute a segmentation function. This is an image whose dark regions are the objects you are trying to
segment.
ii. Compute foreground markers. These are connected blobs of pixels within each of the objects.
iii. Compute background markers. These are pixels that are not part of any object.
iv. Modify the segmentation function so that it only has maxima at the foreground and background marker
locations.
v. Compute the watershed transform of the modified segmentation function.
2.2.3 Improvement of the watershed Transform
In medical images, the markers selection and extraction are not so easy. Some images may be very
noisy and image processing becomes more and more complex. In some cases, the objects to be detected may be
so complex and so varied in shape and size that it is very hard to find improved algorithm enabling their
extraction. For that reason, we need to go a step further in the segmentation approach. We know that the initial
watershed transformation of the gradient image provides very unsatisfactory results many apparently
homogeneous regions are fragmented in small pieces. Fortunately, the watershed transform itself, applied on
another level, will help us to merge the fragmented regions. Indeed, if we look at the boundaries produced by
the segmentation, they do not have the same weight. Those which are inside the almost homogeneous regions
are weaker. In order to compare these boundaries, we need to introduce neighborhood relations between them.
The improved watershed transform presents some advantages: The watershed lines always correspond to the
most significant edges between the markers and it always detects a contour in the area; where there are no strong
edges between the Markers. This contour will be located on the pixels with higher contrast. In this work we first
remove the noise from the image and pixel values are adjusted so that they will help to obtain the well
segmented image.
2.3 Morphological Operations
Mathematical morphology [8, 11] commonly refers to a broad set of image processing operations that
process images based on shapes. Morphological operations selects appropriate structuring element of the
processed image and makes use of the basic theory of morphology including erosion, dilation, opening and
closing operation and the operations of them to get clear MR image edge. In the process, the synthesized modes
of the operations and the feature of structuring element decide the result of the processed image. Detailed
saying, the synthesized mode of the operation reflects the relation between the processed image and original
image, and the selection of structuring element decides the effect and precision and the result. Therefore, the
keys of morphological operations can be generalized for the design of morphological filter structure and the
selection of structuring element. In brain tumor extraction, we select disc shaped structuring element by texture
features of the image and the size, shape and direction of structuring element must been considered roundly. By
the operation features of morphology, erosion and dilation operations satisfy:
𝐅 ⊝ 𝐁 ⊆ 𝐅 ⊆ 𝐅⨁𝐁 𝟏
Where, F denote a gray scale brain image,B denote structuring element.Opening and closing operations satisfy:
𝐅 ∘ 𝐁 ⊆ 𝐅 ⊆ 𝐅 ∙ 𝐁 (𝟐)
In this work morphological operations are applied on the converted binary image. The purpose of the
morphological operators is to extract the tumor part of the MR images.
III. Experimental Results
A large data set consisting of a number of MR tumor images has been collected; all patients have ages
ranging from 18 to 60.Their MRI scans were stored in database of images in JPEG format.

4. Brain Tumor Segmentation and Extraction of MR Images Based on Improved Watershed Transform
DOI: 10.9790/0661-17120105 www.iosrjournals.org 4 | Page
(a) (b) (c)
Fig.2 (a) Original MR Image , (b) Tumor detected as white portion, (c)Brain Tumor extraction
(a) (b) (c)
Fig.3 (a) Original MR Image , (b) Tumor detected as white portion, (c)Brain Tumor extraction
(a) (b) (c)
Fig.4 (a) Original MR Image , (b) Tumor detected as white portion, (c)Brain Tumor extraction
The proposed technique is tested on a large database consisting of 120 tumor images. Fig. (2) to
Fig.(4) shows the tumor extracted from MR brain images. The tumor portion of the MR image is visible, shown
as white color. This portion has the highest intensity than other regions of the image. The algorithm proposed in
this paper is able to extract the brain tumor successfully with 98% accuracy in various age groups. Table 1
shows the comparison of proposed algorithm with [2].
Table 1.Comparision of proposed algorithm
Image set Proposed method Method Proposed in [2]
MR brain tumor Images 118/120 110/120
Success rate(%) 98.33% 91.66%
IV. Conclusions
This paper presented an improvement to the watershed transform for the extraction of brain tumor of
MR images based on segmentation and morphological operator. Moreover, as contrast enhancement is
employed, this technique is able to extract the brain tumors from the MR images in various age groups. The
proposed technique gives more information about brain tumor and helps doctors in diagnosis, the treatment plan

5. Brain Tumor Segmentation and Extraction of MR Images Based on Improved Watershed Transform
DOI: 10.9790/0661-17120105 www.iosrjournals.org 5 | Page
making and state of the tumor monitoring. The experimental results show that the proposed technique can
provide close, smooth and accurate final contours with low computational complexity. The proposed work can
be extended by adapting more segmentation algorithm to suit the different medical image segmentation.
References
[1]. V. Grau, A. U. J. Mewes, M. Alcañiz, Improved Watershed Transform for Medical Image Segmentation Using Prior Information,
IEEE Transactions On Medical Imaging, Vol. 23(4), 2004,447-458.
[2]. A. Mustaqeem, A. Javed, T. Fatima, An Efficient Brain Tumor Detection Algorithm Using Watershed & Thresholding Based
Segmentation, International Journal of Image, Graphics and Signal Processing,Vol.10(3), 2012, 34-39.
[3]. T. Logeswari, M. Karnan, An improved implementation of brain tumor detection using segmentation based on soft computing,
Journal of Cancer Research and Experimental Oncology, Vol. 2(1), 2010, 06-14.
[4]. M. Brummer, R. Mersereau, R. Eisner, and R. Lewine, Automatic detection of brain contours in MRI data sets, IEEE Trans. Med.
Imag.,Vol. 1(4), 1993,153–166.
[5]. M.S. Atkins and B.T. Mackiewich, Fully Automatic Segmentation of the Brain in MRI, IEEE Tran. On medical imaging, Vol. 17
(1 ), 1998, 98-107.
[6]. A. Hammers, Automatic anatomical brain MRI segmentation combining label propagation and decision fusion, NeuroImage, Vol.
33(1), 2006, 115–126.
[7]. Omid Jamshidi and Abdol Hamid Pilevar, Automatic Segmentation of Medical Images using Fuzzy c-Mean and Genetic Algorithm,
Journal of Computational Medicine, Vol. 2(13), 2013,91-96.
[8]. Senthilkumaran N, Kirubakaran C, A Case Study on Mathematical Morphology Segmentation for MRI Brain Image, International
Journal of Computer Science and Information Technologies, Vol. 5 (4), 2014, 5336-5340.
[9]. Hall LO, Bensaid AM, Clarke LP,Velthuizen RP, Silbiger MS, Bezdek JC.1992. A comparison of neural network and fuzzy
clustering techniques in segmenting magnetic resonance images of the brain. IEEE Trans. Neural Networks,Vol. 3(5), 1992, 672-
681.
[10]. R. C. Gonzalez, R.E. Woods and S.L.Eddins, Digital Image Processing Using MATLAB", 2nd Edn., Mc Graw Hill, New
Delhi,2010.
[11]. J.Mehena, Medical Image edge detection based on mathematical morphology, International Journal of Computer and
communication technology, Vol.-2(6), 2011, 45-48.
[12]. M.C. Christ , R. Parvathi, Segmentation of Medical Image Using Clustering and Watershed Algorithms, American Journal of
Applied Sciences,Vol.8(12), 2011, 1349-1352.